PERMAFROST AND PERIGLACIAL PROCESSES Permafrost and Periglac. Process. 19: 31–47 (2008) Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/ppp.609

Solifluction Processes in an Area of Seasonal Ground Freezing, Dovrefjell, Norway

Charles Harris ,1* Martina Kern-Luetschg ,1 Fraser Smith 2 and Ketil Isaksen 3

1 School of Earth Ocean and Planetary Sciences, Cardiff University, Cardiff, UK 2 School of Engineering, University of Dundee, Dundee, UK 3 Norwegian Meteorological Institute, Blindern, Oslo, Norway

ABSTRACT

Continuous monitoring of soil temperatures, frost heave, thaw consolidation, pore water pressures and downslope soil movements are reported from a turf-banked solifluction lobe at Steinhøi, Dovrefjell, Norway from August 2002 to August 2006. Mean annual air temperatures over the monitored period were slightly below 08C, but mean annual ground surface temperatures were around 28C warmer, due to the insulating effects of snow cover. Seasonal frost penetration was highly dependent on snow thickness, and at the monitoring location varied from 30–38 cm over the four years. The shallow annual frost penetration suggests that the site may be close to the limit of active solifluction in this area. Surface solifluction rates over the period 2002–06 ranged from 0.5 cm yr1 at the rear of the lobe tread to 1.6 cm yr1 just behind the lobe front, with corresponding soil transport rates of 6 cm3 cm1 yr1 and 46 cm3 cm1 yr1. Pore water pressure measurements indicated seepage of snowmelt beneath seasonally frozen soil in spring with artesian pressures beneath the confining frozen layer. Soil thawing was associated with surface settlement and downslope soil displacements, but following clearance of the frozen ground, later soil surface settlement was accompanied by retrograde movement. Summer rainfall events caused brief increases in pore pressure, but no further soil movement. Surface displacements exceeded maximum potential frost creep values and it is concluded that gelifluction was an important component of slow near-surface mass movements at this site. Temporal and spatial variations in solifluction rates across the area are likely to be considerable and strongly influenced by snow distribution. Copyright # 2008 John Wiley & Sons, Ltd.

KEY WORDS: solifluction; frost heave; thaw settlement; seasonally frozen ground; pore pressures

INTRODUCTION of soil temperatures, frost heave, thaw consolidation, pore water pressures and downslope soil movement In this paper we report field measurements of solifluc- has allowed us to analyse the mass movement tion at Steinhøi, Dovrefjell, Norway (latitude 628 150 processes associated with soil freezing and thawing. 2800 N, longitude 098 510 5800 E, altitude 1396 m, An instrument package was developed for this Figure 1), where seasonal ground freezing and purpose during full-scale laboratory simulation thawing are from the surface downwards and where studies (Harris et al., 1996), and an equivalent permafrost is absent. Detailed continuous monitoring weather-proofed version was installed on a prominent solifluction lobe at Steinhøi in August 2001 (see Harris et al., 2007, for details). A lightning strike in * Correspondence to: Charles Harris, School of Earth Ocean and Planetary Sciences, Cardiff University, Cardiff CF10 3YE, November 2001 caused instrument failure which UK. E-mail: [email protected] was not repaired until 11 July 2002, and therefore data Received 17 September 2007 Revised 18 December 2007 Copyright # 2008 John Wiley & Sons, Ltd. Accepted 18 December 2007 32 C. Harris et al.

Figure 1 Location of the Steinhøi solifluction monitoring station.

only from the four year period from summer 2002 to The role of high soil moisture content in promoting summer 2006 are presented. The study is focused on gelifluction has been emphasised in many field studies processes rather than regional soil displacement rates. (e.g. Benedict, 1970; Washburn, 1979; Harris, 1981; Given the range of surface gradients, snow conditions, Jaesche et al., 2003; Kinnard and Lewkowicz, 2005) drainage, etc., within the Steinhøi area, coupled with and several authors have reported moisture contents annual climatic variability, it is likely that rates of close to the Liquid Limit (e.g. Washburn, 1967; Smith, solifluction are spatially and temporally highly 1988; Kinnard and Lewkowicz, 2005). During variable (see for instance, Washburn, 1999). laboratory simulations Harris et al. (1995, 1997, Although Andersson (1906) did not exclude non- 2003, 2008) demonstrated very low shear strengths periglacial environments in his original definition, the that persisted for only short periods immediately term solifluction has been widely used to describe behind the thaw front when void ratios were high. slow periglacial mass movement caused by frost creep Subsequent thaw consolidation rapidly lowered and gelifluction (e.g. Ballantyne and Harris, 1994; both void ratio and moisture content, and led to a French, 2007; Matsuoka, 1998, 2001). Gelifluction rapid increase in soil strength. The mobilised and frost creep apparently operate in tandem, since frictional strength of these thawing ice-rich soils both result directly from frost heave and thaw was largely a function of pore water pressure. High settlement. Where freezing and thawing are shallow pore pressures and low effective stress values were and of short duration, creep movements dominate (e.g. recorded in the laboratory experiments for short Matsuoka, 2005), while gelifluction occurs during periods (often only a few hours) immediately behind seasonal thawing of ice-rich soil (Washburn, 1967; the thaw front. Harris and Davies, 2000; Harris, 2007; Matsuoka, Continuous monitoring in the Alps allowed Jaesche 2001). Recent laboratory studies have shown that et al. (2003) to observe soil movements that lasted gelifluction is best considered as elasto-plastic several days at given depths in a thawing soil, while deformation that might be described as accelerated Kinnard and Lewkowicz (2005) in southwest Yukon pre-failure creep rather than viscous flow (Harris observed repeated discrete periods of displacement et al., 2003). rather than slow incremental strains. Similarly,

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp Solifluction Processes, Dovrefjell, Norway 33

Matsumoto and Ishikawa (2002) demonstrated that, on mantled by Weichselian till with clasts up to boulder a gelifluction lobe in Sweden, soil movement was size set in a frost-susceptible matrix of silt and concentrated in two periods when ice-rich parts of the soil fine sand. Solifluction processes have reworked profile, firstly near the surface and secondly at a depth of the till, and excavation during installation of instru- around 20 cm, were in the process of thawing. mentation some 15 m upslope from the lobe front The aim of the present paper is to describe con- revealed 1–1.2 m of reworked till overlying bedrock tinuous field data collected from a slope where and containing thin organic layers that likely extend solifluction processes are active, in order that the from the lobe front and represent palaeo-surfaces influence of soil thermal conditions, frost heave/thaw buried by frontal advance (e.g. Matthews et al., 1986, settlement and hydraulic conditions may be explored. 2005). Soil grain size distribution was measured from five samples collected at 10 cm depth intervals SITE DESCRIPTION to a maximum depth of 60 cm. Clay contents were less than 4% except for the sample from The site lies above the treeline on the southern flanks 50–60 cm which showed 13%. Silt contents rang- of Steinhøi (Figure 1). It consists of a well-developed ed from 16% to 38%. Occasional large clasts were not turf-banked solifluction lobe with a tread length of included in grain-size determinations. The soil has low approximately 30 m and riser height around 1.5 m. plasticity (Plasticity Index 5%) with Plastic Limit 27% (Figure 2). Turf-banked lobes are widespread across and Liquid Limit 32%. this slope. Bedrock comprises grey Proterozoic and The nearest meteorological station at Fokstugu Cambrian phyllite, biotite-phyllite and shale and (972 m), on Dovrefjell, has a mean annual air forms part of the early Ordovician Gula-nappe temperature of 0.18C (1961–90) and mean annual (Geological Survey of Norway, 2002). The area is precipitation of 435 mm. Hourly air temperatures

Figure 2 The Steinhøi monitored solifluction lobe showing the frost heave frame, with the LVDT fixed base triangle in the foreground and logger box and solar panel in the background. The apex of the LVDT triangle is attached to a footplate embedded in the ground surface (see Harris et al., 2007, for details). Thermistors measuring air temperatures are shielded with white plastic tubing and attached to the solar panel pole. The station is fenced against large animals.

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp 34 C. Harris et al. measured 2 m above the ground surface at Steinhøi of 80 cm (Figure 3). Thermistors, shielded by 2 cm during 2003, 2004 and 2005 show annual means of diameter white plastic tubing, were also mounted on 0.18C, 0.88C and 0.18C, but mean ground the pole supporting the solar panel at heights of 50, surface temperatures were more than 28C higher, 100 and 200 cm. Temperature measurement errors are due largely to the insulating effect of winter snow. estimated to be up to 0.48C. The 10 cm thermistor Storms are common in winter, accompanied by strong proved unreliable, and data from this depth were not winds, and redistribution of snow by deflation leads to used. considerable spatial variation in snow thickness. Pore Water Pressure

INSTRUMENTATION A vertical array of five Druck PDCR81 pore pressure transducers was installed at 10 cm depth intervals Instrumentation was designed to monitor three key between 20 and 60 cm. Transducers were filled with elements: ground surface movements (frost heave, low viscosity silicon oil and had a range of 350 mb, thaw settlement and downslope solifluction), ground combined non-linearity and hysteresis of 0.2% and and air temperatures, and pore water pressures. In thermal sensitivity of 0.2% of reading per 8C. addition, profiles of soil movement (the distribution of shear strain with depth) were measured over a four Ground Surface Movements year period (2002–06). Continuous monitoring of ground surface movements Ground and Air Temperatures requires a stable reference datum, though Berthling et al. (2000) reported three-dimensional target move- Stainless steel thermistor probes supplied by Camp- ments measured using carrier-phase Differential bell Scientific Ltd formed a vertical array above and Global Positioning System (DGPS). In the present below the ground surface. Thermistors were installed study, a stable frame was constructed from steel at the ground surface and at 10 cm intervals to a depth scaffolding poles of external diameter 5 cm. The frame

Figure 3 Site plan showing location of instrumentation.

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp Solifluction Processes, Dovrefjell, Norway 35 dimensions were 2 m 2.2 m 0.95 m. Diagonal side sections of plastic tubing of external diameter 2 cm. struts increased rigidity. The supporting legs were Sections were threaded over a steel rod slightly vertical, though subsequent laboratory physical smaller in diameter than the internal diameter of the modelling has shown the development of significant tubing to form column lengths of around 75 cm, downslope bending stresses as a result of winter frost and these were pushed into an auger hole perpen- heave perpendicular to the surface (Ripley, 2004). dicular to the ground surface. The steel rod was then Frame legs extended to bedrock at a depth of 1–1.2 m withdrawn, leaving column sections free to move with and the lowermost 0.5 m was embedded in concrete. the soil. Columns were installed in July 2001 and The remaining buried section was greased and re-excavated in August 2005, indicating the accumu- sheathed in telescopic plastic tubing to isolate it from lated movement over a four year period. adjacent frost heaving soil. Soil surface motion was measured using a pair of waterproof captive guided armature DC/DC linear RESULTS variable differential transformers (LVDTs) (350 mm range). The configuration of LVDTs allowed measure- Thermal Regime ment of soil displacement with an accuracy of 1.5 mm over the operating temperature range. Air Temperatures and Inferred Snow Depths. Two LVDTs were mounted via spherical end bearings Figure 4 shows temperatures recorded at 200, 100, on a central scaffolding pole supported by the frame, 50 and 0 cm above the ground surface from 2002 to running parallel to the ground surface along the slope 2006. Minimum temperatures of around 208C were fall line and extending upslope from the frame recorded, but only January was free of above zero (Figure 2). LVDTs formed a fixed base triangle at air temperatures in each of the four years. The the apex of which was an 8 cm 5 cm stainless steel temperature record shows that during winter 2002–03, footplate embedded in the soil surface. Frost heave, the thermistor probe located 50 cm above the ground thaw settlement and downslope displacements of the surface lay above the snow (time bar (a), Figure 4) ground surface were detected by changes in length of since its temperature time series was virtually the LVDTs and hence geometry of the fixed base identical to those at 100 and 200 cm. In 2003–04 triangle (see Harris et al., 2007, for details). temperature fluctuations in the 50 cm thermistor were The horizontal cross members supporting the LVDT significantly dampened from early February until late beam were instrumented with pairs of strain gauges April (time bar (b), Figure 4), indicating that snow glued to the steel poles with epoxy resin. Although not depth was greater than 50 cm but less than 100 cm in directly calibrated, these indicated periods when snow this period. In 2004–05 snow depth was greater than loading caused slight bending and therefore lowering 50 cm from mid January to early June (time bar (c), the LVDT support beam relative to the ground surface, Figure 4) and greater than 100 cm through much of registering as apparent heave in the LVDT readings. April and May (time bar (d), Figure 4). During this Such loading was evident in the winter of 2004–05, period of thick snow cover the frame suffered signifi- and LVDT data from that winter are not presented cant snow loading, and LVDT data could not be used here. In other years no significant deflection of the to determine ground surface movements. Winter beams was apparent. 2005–06 was characterised by repeated short-term surface freeze-thaw cycles in the autumn (Table 1), Data Acquisition (time bar (e), Figure 4) with final winter surface freezing not commencing until 7 November 2005, All data were recorded at hourly intervals via some three to four weeks later than in the previous Campbell Scientific CR23X loggers with a 16/32 three years. Snow thickness was less than 50 cm until channel multiplexer. The logger was powered by a mid March but from mid March until the beginning of 12 V heavy-duty car battery charged with an 18 W May, depths were between 50 and 100 cm (time bar solar panel (Figure 2). All cabling between sensors (f), Figure 4). and logger was either buried, or protected from damage by animals by plastic tubing. Soil Temperatures. The near-surface ground thermal regime is strongly Profiles of Soil Movement: Rudberg Columns influenced by snow conditions, with the relatively snow-poor winter of 2002–03 showing greatest frost Three Rudberg columns were installed in a downslope penetration (38 cm), compared with around 30 cm in transect (Figure 3). Columns comprised 2 cm long 2003–04 and 2004–05, and slightly more than 30 cm

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp 36 C. Harris et al.

Figure 4 Air and ground surface temperatures from August 2002 to July 2006. Time bars are referred to in the text. in 2005–06 (Figure 5). Ground surface temperatures variation in pore pressure (Figure 7c and 7d). Pressures were lowest in 2002–03, with minimum recorded were low but fluctuating through the winter, and temperature 48C in December (Figure 5). increased very rapidly during spring as air temperatures Thermistors located below the depth of seasonal rose and the nival thaw commenced, confirming that frost penetration recorded strong cooling in late snowmelt was flowing beneath the frozen near-surface winter/early spring (indicated by time bars (a)–(d) in soil layers. In 2004 this seepage began at 40 cm depth Figure 6). Cooling was caused by heat advection as some three weeks before the soil surface temperatures cold meltwater from higher upslope flowed beneath rose above zero, and in 2005 and 2006 some four weeks the frozen surface layer. The effect was greatest in the before surface thaw commenced. Pore pressures were in snow-rich winter of 2004–05 (time bar (c), Figure 6), excess of hydrostatic, particularly during the period and least in the snow-poor winter of 2002–03 (time bar when the still-frozen near-surface soil formed a (a), Figure 6). confining layer. At 40 cm, pore pressures reached a maximum of 96 cm of water (9.8 kPa) for a few hours in Pore Water Pressures late April 2004 (Figure 7c), which is in excess of geostatic (meaning the self weight of the overburden Figure 7 presents pore pressures recorded at 20, 40 and would have been supported by pore pressures). 50 cm depths, together with soil temperatures. The However, at that time the frozen ground above was transducers at 40 and 50 cm depths remained unfrozen sufficiently rigid to prevent shallow landsliding. By the throughout the period and showed considerable time the surface layer had thawed, pore pressures at 40

Table 1 Summary of time periods with frozen ground.

2002–03 2003–04 2004–05 2005–06 Autumn frost cycles at 0 cm 2 2 0 6 Winter ground freezing begins 07/10/02 19/10/03 11/10/04 7/11/05 Completion of ground ice melt 06/06/03 26/05/04 04/07/05 12/06/06

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp Solifluction Processes, Dovrefjell, Norway 37

Figure 5 Isotherms in degrees Celsius showing frost penetration over four winters, from 2002 to 2006. Shaded areas are below zero Celsius.

and 50 cm depth were only slightly in excess of may indicate a zone of preferential seepage, probably hydrostatic and insufficient to cause instability. At relating to higher permeability at this depth. The rapid 50 cm, pore pressures were slightly lower, rising to fall in pressures in late spring or early summer was above 60 cm water (5.9 kPa) in spring 2004 and 2005, synchronous at 40 and 50 cm in all three years, and to a maximum of 45 cm water (4.4 kPa) in spring reflecting the final clearance of much of the snow 2006, and remaining high for a further three to four upslope from the site and therefore the loss of melt- weeks after soil thawing was complete. water seepage. Temporal periodicity in pore pressure The fact that pore pressures at 40 cm were higher fluctuations was generally longer than diurnal (though than at 50 cm and began to rise before those at 50 cm a weak diurnal signal is also detectable) prior to

Figure 6 Soil temperatures 2002–06. Time bars indicate periods of late winter/early spring ground cooling caused by meltwater flow beneath the frozen near-surface layers.

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp 38 C. Harris et al.

Figure 7 Time series of soil temperatures and pore water pressure. (a) Soil temperatures. (b) Pore water pressures recorded at 20 cm depth. Pore pressure response to the summer rainfall event in August 2005 is labelled. (c) Soil temperatures and pore water pressures recorded at 40 cm depth. Pore pressure responses to summer rainfall events are labelled. (d) Pore water pressures recorded at 50 cm depth. Pore pressure responses to summer rainfall events are labelled.

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp Solifluction Processes, Dovrefjell, Norway 39 clearance of the frozen surface layer, but subsequently the first two weeks of May 2004 (Figures 7b and 8), became markedly diurnal (Figure 8). Occasional when ice pressures increased rapidly and renewed influxes of surface water associated with rainfall frost heave was registered by the LVDTs (see below). events were registered through the summer periods This is probably equivalent to Mackay’s observation when the soil was completely thawed (examples are of ‘summer heave’ above permafrost in the Canadian labelled in Figure 7c and 7d). arctic (Mackay, 1983), though in this seasonally At 20 cm depth, the pore pressure transducer frozen context, resulting soil ice soon thawed again. response was related closely to phase changes within Ice pressures fell rapidly as thawing at 20 cm began. the soil in winter and spring, though the signal from Thawing was complete on 26 May 2004, when pore 2005 is less clear than in the previous and succeeding pressures at 20 cm rose to slightly in excess of years (Figure 7b). All transducers were recalibrated hydrostatic (equivalent to 23.6 cm water or 2.3 kPa), and re-installed in July 2005. During winter 2003–04 and fluctuated around this level for the next 20 days. the minimum temperature recorded at 20 cm depth Thus the soil remained saturated at 20 cm depth for was 0.28C, and sub zero temperatures persisted from around three weeks after thawing in 2004, correspond- early December to late May. It should be noted that the ing to the period of high pore pressures recorded at precision of temperature measurements is insufficient 40 cm depth. to discriminate very small temperature changes, The winter of 2004–05 was snow-rich and initiation but evidence of frost heave and measurement of of ground thawing was delayed until late June. heaving pressures provide clear evidence of freezing Consistent ice pressures were not recorded at 20 cm at this depth. The formation of segregation ice led (Figure 7b), but immediately following soil thaw, to significant ice pressures in winter 2003–04 pore pressures rose to a maximum value of 28.5 cm (Figure 7b). These declined slowly as heaving rates water (27.9 kPa) and remained elevated for three decreased. Surface thaw began on 17 April 2004, and weeks. The rainstorm event in the second week of surface temperatures rose rapidly on 7 May. The thaw August 2005 was clearly registered, with pore front reached a depth of 20 cm on 18 May. Prior to pressures of 20.8 cm water (labelled in Figure 7b), this, meltwater from the surface apparently migrated equivalent to a water table at or slightly above the downwards into the underlying still-frozen soil (along ground surface. the thermal gradient) during thaw penetration and In winter 2005–06, winter ice segregation pressures caused renewed ice segregation at 20 cm depth during were again registered at 20 cm, though not until

Figure 8 Air temperature, soil temperature at 20 cm and pore pressures, spring 2004. The bar labelled WP indicates the period when pore water pressures (as opposed to ice pressure) were recorded at 20 cm depth. The bar labelled M shows the period when soil surface downslope displacements took place.

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp 40 C. Harris et al.

March. These persisted until initiation of thawing at repaired in August 2005, so that data for 2005–06 are this depth, on 10 June 2006, when ice pressures fell re-zeroed and presented separately. rapidly and soil temperatures rose sharply above zero Movement of the LVDT footplate was resolved into (Figure 7b). A small peak in ice pressures immediately displacements perpendicular to the ground surface before thawing probably corresponded to ice segre- (heave and settlement) and parallel to the ground gation due to downward migrating water, and this was surface (downslope movement and retrograde move- registered by the LVDTs as slight surface frost heave ment). (see below). Following thaw, pore pressures at 20 cm were again in excess of hydrostatic, decreasing over a Phases of Heave-settlement through Time. period of some two weeks from an initial high of Five distinct periods may be identified in the record around 33 cm water (3.4 kPa). for 2002–03 and 2003–04, labelled (a)–(e) in Figure 9. In the summer (period (a)), soils were relatively dry Frost Heave, Thaw Settlement and Downslope and the ground largely stable. Period (b), in autumn Surface Displacements and winter, showed frost heave and apparent upslope movement of the footplate during ground freezing. Data were collected for the period from August 2002 Heaving was initially rapid, but slowed later in the to August 2006, but thick snow cover in winter 2004– winter. Apparent upslope movement of the footplate is 05 caused deformation of the beam supporting the ascribed to small downslope movements of the LVDT LVDTs so that apart from the early heaving phase support bar resulting from lateral stresses on the prior to the commencement of snow loading, data for vertical frame legs caused by frost heave perpendicu- this year cannot be interpreted. The record shows that lar to the ground surface (see Harris et al., 2007, for both frame and LVDTs were also disturbed in July details). The first sign of thaw settlement marked the 2005, possibly by a large animal. The system was start of period (c) when ground surface temperatures

Figure 9 Heave, settlement and downslope/upslope soil movement (left axis) plotted with temperatures (right axis) August 2002–July 2004. Upper graph shows temperatures at the surface and 20 cm. Lower graph shows LVDT movement data with grey line marking heave and resettlement and black line downslope (increasing values) and upslope (decreasing values) soil movements. See text for explanation of time periods labelled (a) to (e).

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp Solifluction Processes, Dovrefjell, Norway 41

Figure 10 Phases of heave, settlement and downslope/upslope soil movement (left axis) plotted with temperatures (right axis), August 2005–August 2006. Upper graph shows temperatures at the ground surface and at 20 cm. Lower graph shows LVDT movement data. Grey line — heave and resettlement, black line — soil movements downslope (increasing values) and upslope (decreasing values). See text for explanation of time periods labelled (a) to (e). were 08C and thaw had begun. In all years, but decrease in the rate of settlement and the development particularly in 2004, initial thaw settlement and of retrograde (upslope) movement of the soil surface accompanying downslope surface movements in relative to the frame. The LVDT mounting beam was period (c) were interrupted by slight heaving towards stable at this stage since there were no longer any the end of the period. This renewed heaving lateral heaving pressures on the frame legs. The corresponded with a sharp increase in ice pressures upslope surface ground movement represented retro- at 20 cm, a few days before the thaw plane arrived grade movement as the soil dried (e.g. Washburn, (see above). 1967; Harris and Davies, 2000). The establishment of a steeper thermal gradient Figure 10 shows a similar pattern of frost heave, thaw in the thawing soil as ground surface temperatures settlement and downslope surface movements in rose, resulted in more rapid thaw penetration with 2005–06 compared with the earlier years, though with associated thaw settlement and downslope surface some variations. At the end of period (a), in the last movements (period (d)) and corresponded with the week of October 2005, a cycle of surface freezing and main phase of ground thawing. Period (d) lasted thawing caused initial heave and thaw settlement before 19 days (26 May–14 June) in 2003 and 13 days the main winter phase of soil freezing commenced (8–21 May) in 2004, and both settlement and (period (b), Figure 10). Frost heaving through the downslope surface displacements were rapid. The winter was progressive, but was less than in 2002–03 or final stage in the annual cycle (period (e)) followed 2003–04 (Table 2). The soil surface temperature clearance of ground ice, and corresponded to a rose to 08C on 1 May 2006, and immediate slight

Table 2 Frost heave, thaw settlement and downslope surface displacements, 2002–03, 2003–04 and 2005–06.

2002–03 2003–04 2005–06 Net frost heave (mm) 38 44 29 Net thaw settlement (mm) 34 40 25 Net surface downslope movement (mm) 13 13 9

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp 42 C. Harris et al.

Figure 11 Vectors of surface movement: (a) 2002–03 and 2003–04; (b) 2005–06.

thaw settlement was registered (initial part of period surface also apparently rose by a few millimetres each (c)). This was followed by heave amounting to some year relative to the frame. This apparent rise was due 7 mm, then further and much greater thaw settlement, to a progressive decrease in the ground surface slope, with the whole cycle being complete by 8 May. No ice from approximately 148 to 128, and the fact that the pressures were registered at 20 cm during the thaw slope of the frame beam supporting the LVDT triangle phase, so it is likely that thaw-phase heaving was due to was slightly greater than that of the underlying ground, ice segregation at shallower depths for a period of two so that as the footplate moved downslope, the height or three days following initiation of surface thaw. difference between the support beam and ground Despite surface temperatures rising above 08Catthe surface decreased slightly. Net heave and downslope end of May, the onset of rapid thaw settlement and displacement data are presented in Table 2. Although downslope displacement (period (d)) was delayed until heave is apparently the dominant factor in the amount 7 June, movement continuing for only four days until of soil movement, greater retrograde movement in 10 June. Period (d) was followed, as previously, by a 2004 resulted in a smaller net displacement than in phase of further settlement and retrograde movement 2003, despite greater frost heave. (period (e)) at the end of the heave-settlement cycle.

Vectors of Soil Surface Movement. Profiles of Soil Movement Combining heave and downslope displacement data provides a vector of soil surface movement through Three Rudberg columns, located along the lobe axis, each annual cycle (Figure 11). The apparent upslope were excavated in summer 2006 (Figure 12) and movement during heave caused by deflection of the revealed net soil movement profiles over a period of frame is clear in all three years. The short periods four years. Average annual surface displacement, the of renewed ice segregation caused by refreezing of depth of displacement and the volumetric transport percolating meltwater following initiation of surface rate increased significantly from the rear of the lobe thaw are indicated in Figure 11. These were followed tread to the frontal zone (Table 3). In the higher part of by a phase of gelifluction during thaw consolidation, the lobe tread, soil displacements were detected to a and then a period of retrograde movement associated depth of 22 cm in profile 1, while in the mid-tread with soil drying and consolidation following final location, just upslope of the frame, the depth of thaw. In each year the LVDT footplate embedded in movement was 43 cm (profile 2), and in the lowermost the ground surface moved downslope, but the ground measured displacement profile, below the monitoring

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp Solifluction Processes, Dovrefjell, Norway 43

Figure 12 Soil displacement profiles August 2001–August 2005. (a) Measured profiles; (b) profile 1 (rear of lobe tread); (c) profile 2 (immediately above monitoring frame); (d) profile 3 (below monitoring frame in lobe frontal zone).

Table 3 Soil displacement measured by Rudberg columns installed in August 2001 and excavated in August 2005.

Profile 1 Profile 2 Profile 3 Average surface movement rate (cm yr1) 0.5 1.0 1.6 Average volumetric transport rate (cm3 cm1 yr1) 6 22 46 Maximum depth of soil movement (cm) 22 43 50

frame, this depth of displacement reached 50 cm. (Figure 11). Once thawing was complete, downslope Movement rates decreased with depth in profiles 1 and displacements in 2004 and 2006 ceased, but in 2003 2, but in profile 3 a convexo-concave profile was settlement and downslope movement continued for a observed, similar to that described by Matsuoka further eight days. Thus the period of downslope (2001) as typical of solifluction profiles caused by surface movement was relatively short, lasting one-sided soil freezing and thawing. Typical rates of approximately 19 days in 2003, 13 days in 2004 soil transport by solifluction are quoted by Matsuoka and only four days in 2006. Measured average rates of (2001) as being in the range 20–50 cm3 cm1 yr1, movement during these periods were 1.1 mm per day which corresponds well with observations here. in 2003, 1.5 mm per day in 2004 and 2.5 mm per day in 2006. Washburn (1967) defined the term potential frost DISCUSSION creep as the maximum creep that would be anticipated arising from frost heave perpendicular to the ground Mechanisms of Movement surface, followed by vertical settlement during soil thaw, under the influence of gravity. Potential frost Despite the extended periods of soil saturation creep is therefore the product of total heave and the following ground thawing and short-term periods of tangent of the slope angle. It is generally considered raised pore pressures during summer rainstorms, that actual creep movements are less than this downslope soil displacements were recorded only potential value due to the effects of cohesion within during the period of thaw consolidation. Later surface the moist soil mass. Figure 13 presents an analysis of settlement was associated with retrograde movement the main thaw settlement phases in each of the three

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp 44 C. Harris et al.

Figure 13 Vectors of surface movement during the main periods of thaw settlement in 2003, 2004 and 2006. Dashed arrows show the trajectories of the soil surface if thaw settlement were vertical, that is, the potential frost creep trajectories. For labelling of the 2003 vector, see text.

years 2003, 2004 and 2006. It is assumed that the 12.7 mm). However, in 2006 the downslope displace- slight downslope distortion of the frost heave frame ment arising from the main phase of thaw settlement (manifest in apparent upslope component of frost totalled only 5.2 mm, compared with an annual heave in the observed surface vectors — see measured total of 9 mm from the vector in Figure 11) was progressively recovered through this Figure 11b. The discrepancy is due to an additional period of thaw settlement, and that the frame returned increment of movement resulting from the short to its pre-frozen geometry by the end of the thaw period of soil freezing and thawing in the autumn of cycle. This has been accounted for in Figure 13 by 2005 which lasted from 22 October to 10 November, distributing the additional apparent ground surface prior to the main winter cycle of freezing. movement due to frame recovery evenly through the Two main phases of thaw settlement are recognised: thaw period, and subtracting this from the raw data. firstly, settlement associated with gelifluction/creep, and Potential frost creep accounted for 66% of the net secondly a succeeding period of settlement associated surface movement (including the retrograde move- with retrograde movement. The gelifluction/creep phase ment) in 2003, 53% in 2004 and 76% in 2006. As in 2003 was subdivided into two periods of more rapid Figure 13 demonstrates, it is, in reality, impossible to downslope movement labelled (a) and (c) in Figure 13, separate creep movements from gelifluction. and a period with much less downslope movement, Total downslope movements in 2003 and 2004 labelled (b). There appears to be some consistency in the shown in Figure 13 are identical to those determined ratio of downslope surface displacement to thaw for the full vectors in Figure 11a (i.e. 13.4 and settlement during the gelifluction/creep stage over the

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp Solifluction Processes, Dovrefjell, Norway 45 monitored three years. With the exception of period (b) variables. Through this approach we have explored the (Figure 13) in 2003, when it was 0.13, the ratios in relationships between soil movements and a range of 2003 were 1.05 and 0.74 (periods (a) and (b)), factors including ground temperatures, frost heave, respectively. In 2004 the ratio was 0.8 and in 2006 it thaw settlement and soil moisture conditions, particu- was 1.04. The equivalent ratio for vertical soil larly pore pressures. resettlement (maximum potential frost creep) is around Observed rates of solifluction on the monitored 0.23 (the tangent of 138). solifluction lobe at Steinhøi, Norway, were lowest near Harris et al. (1997) demonstrated a consistent the rear of the lobe (0.5 cm per year over a four year displacement-settlement ratio over a number of period, equivalent to 6 cm3 cm1 yr1 volumetric soil freeze-thaw cycles of around 1.5 for a silt-rich test transport) and highest near the front (1.6 cm yr1, soil used in full-scale laboratory simulations on a equivalent to 46 cm3 cm1syr1 volumetric soil slope with a gradient of 128 (similar to the gradient transport). Continuous monitoring has shown that here). Thus, it appears that the laboratory soil had a following initiation of surface thaw, downward higher susceptibility to solifluction than the present percolation of meltwater into the still-frozen soil Steinhøi field soil. below caused renewed ice segregation and slight frost heave. Subsequently, soil thawing was associated with settlement and downslope displacements, the main period of solifluction lasting 19 days in 2003, 13 days Pore Water Pressures in 2004 and only four days in 2006. Thus, it is concluded that soil shear strain occurred immediately Simulation experiments, both at full-scale (Harris behind the advancing thaw front as a result of high et al., 1997) and at reduced scale in the geotechnical pore pressures generated as water-filled voids were centrifuge (Harris et al., 2003, 2008), have demon- closed during thaw consolidation. At 20 cm, however, strated the importance of low soil shear strengths recorded pore pressures were also affected by melt- during gelifluction. High pore pressures have been water seepage at greater depths. recorded during thaw consolidation when both soil Frost penetration was shown to be highly sensitive void ratios and moisture contents are initially high, but to snow depth and ranged from 38 cm in 2002–03 decreasing. Here we recorded maximum pore press- when snow depth was less than 50 cm to around 30 cm ures of between 3.4 kPa and 2.6 kPa during thaw during 2004–05 when maximum snow depths exce- consolidation at 20 cm depth, exactly within the range eded 100 cm. Large spatial variation in frost penetra- of values recorded at similar depths in the full-scale tion depths in response to variable snow thickness is and centrifuge simulation experiments referred to also suggested by meltwater seepage beneath the above. At some point during consolidation, void seasonally frozen surface layer of the monitored ratios and moisture contents reach critical values at solifluction lobe. It is assumed that snow arrived which the soil becomes too stiff to undergo further sufficiently early and was sufficiently deep to prevent shear strain under self weight stresses. Gelifluction ground freezing in certain accumulation hollows then ceases, though volume strain continues due to upslope from the monitored lobe, thus providing soil drying. At Steinhøi, post-thaw pore water entry points for groundwater seepage beneath the pressures at 20 cm were modulated strongly by high seasonally frozen layer further downslope. In pore pressures associated with meltwater seepage at 2005–06, frost heave and thaw settlement were much 40 cm depth, and therefore the thaw consolidation less than in previous years, and since the ratio of signal is less clear than might otherwise have been surface gelifluction to thaw settlement was roughly the case. constant in each year, the amount of gelifluction was also significantly lower. It is concluded that, as observed by Jaesche et al. (2003) in the Austrian Alps, spatial and temporal variability in snow cover CONCLUSIONS probably plays the greatest role in determining frost penetration depths and frost heave totals, and This paper, like several recent contributions to our therefore contributes significantly to spatial and understanding of periglacial solifluction processes temporal variability in downslope soil transport by (e.g. Kinnard and Lewkowicz, 2005; Jaesche et al., gelifluction. Any future trends toward warmer 2003; Matsumoto and Ishikawa, 2002; Matsumoto winter temperatures or increased snow fall at this et al., 2001; Matsuoka, 1994, 2005), has emphasised site would have the effect of reducing rates of the importance of continuous monitoring of site gelifluction.

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp 46 C. Harris et al.

ACKNOWLEDGEMENTS Harris C, Davies MCR, Coutard J-P. 1997. Rates and processes of periglacial solifluction: an experimental The authors wish to thank the following for their approach. Earth Surface Processes and Landforms 22. support in the field: Professor Johan Ludvig Sollid, Harris C, Davies MCR, Rea BR. 2003. Gelifluction: James Smith, Rune Strand Ødega˚rd, Karsten Vedel viscous flow or plastic creep? Earth Surface Processes Johansen and Trond Eiken. The mountain supervisor and Landforms 28: 1289–1301. Harris C, Luetschg MA, Davies MCR, Smith FW, Chris- in Folldal, Mr Odd Enget, gave permission for the tiansen HH, Isaksen K. 2007. Field instrumentation for installations at Steinhøi and helped with local trans- real-time monitoring of periglacial solifluction. Per- port. Funding for this research was provided by the mafrost and Periglacial Processes 18: 105–114. DOI: British Engineering and Physical Sciences Research 10.1002/ppp.573 Council (EPSRC GR/T22964) and by Cardiff Univer- Harris C, Smith JS, Davies MCR, Rea B. 2008. sity. Logistical support from the University of Oslo An investigation of periglacial slope stability in and the Norwegian Meteorological Institute is grate- relation to soil properties based on physical modelling fully acknowledged. Professor J.L. Sollid kindly in the geotechnical centrifuge. Geomorphology 93: allowed us to use the facilities of the Hjerkinn field 437–459. doi:10.1016/j.geomorph.2007.03.009. station. The comments of the anonymous reviewers on Jaesche P, Veit H, Huwe B. 2003. Snow cover and soil moisture controls on solifluction in an area of seasonal an earlier version of the paper are appreciated. frost, Eastern Alps. Permafrost and Periglacial Pro- cesses 14: 399–410. DOI: 10.1002/ppp.471 Kinnard C, Lewkowicz AG. 2005. Movement, moisture and thermal conditions at a turf-banked solifluction REFERENCES lobe, Kluane Range, Yukon Territory, Canada. Perma- frost and Periglacial Processes 16: 261–275. DOI: Andersson JG. 1906. Solifluction, a component of sub- 10.1002/ppp.530 aerial denudation. Journal of Geology 14: 91–112. Mackay JR. 1983. Downward water movement into Ballantyne CK, Harris C. 1994. The Periglaciation of frozen ground, western Arctic coast, Canada. Cana- Great Britain. Cambridge University Press: Cam- dian Journal of Earth Sciences 20: 120–134. bridge. Matsumoto H, Ishikawa M. 2002. Gelifluction within Benedict JB. 1970. Downslope soil movement in a Color- a solifluction lobe in the Ka¨rkevagge valley, ado alpine region: rates, processes, and climatic Swedish Lapland. Geografiska Annaler 84A: 261– significance. Arctic and Alpine Research 2: 165–226. 265. Berthling I, Eiken T, Sollid JL. 2000. Continuous Matsumoto H, Kurashige Y, Hirakawa K. 2001. Soil measurements of solifluction using carrier-phase moisture conditions during thawing on a slope in differential GPS. Norsk Geografisk Tidsskrift- the Daisetsu Mountains, Hokkaido, Japan. Permafrost Norwegian Journal of Geography 54: 182–185. and Periglacial Processes 12: 211–218. DOI: 10.1002/ French HM. 2007. The Periglacial Environment,3rd edn. ppp.364 John Wiley: Chichester. Matsuoka N. 1994. Continuous recording of frost heave Geological Survey of Norway. 2002. Bedrock map and creep on a Japanese high mountain slope. Arctic Folldal 1519 II, Scale 1:50 000. and Alpine Research 26: 245–254. Harris C. 1981. Periglacial Mass Wasting: A Review of Matsuoka N. 1998. The relationship between frost heave Research. BGRG Research Monograph. Geo Books: and downslope soil movement: field measurements in Norwich. the Japanese Alps. Permafrost and Periglacial Pro- Harris C. 2007. Slope deposits and forms. In Encyclo- cesses 9: 121–133. pedia of Quaternary Science, Elias SA (ed.). Elsevier: Matsuoka N. 2001. Solifluction rates, processes and Amsterdam; 3, 2207–2216. landforms: a global review. Earth-Science Reviews Harris C, Davies MCR. 2000. Gelifluction: observations 55: 107–133. from large-scale laboratory simulations. Arctic, Ant- Matsuoka N. 2005. Temporal and spatial variations in arctic and Alpine Research 32(2): 202–207. periglacial soil movements on alpine crest slopes. Harris C, Davies MCR, Coutard J-P. 1995. Laboratory Earth Surface Processes and Landforms 30: 41– simulation of periglacial solifluction: significance of 58. porewater pressure, moisture contents and undrained Matthews JA, Harris C, Ballantyne CK. 1986. Studies shear strength during thawing. Permafrost and Peri- on a gelifluction lobe, Jotunheimen, Norway: 14C glacial Processes 6: 293–312. chronology, stratigraphy, sedimentology and palaeoen- Harris C, Davies MCR, Coutard J.-P. 1996. An exper- vironment. Geografiska Annaler 86A: 345–360. imental design for large-scale modelling of solifluction Matthews JA, Seppa¨la¨ M, Dresser PQ. 2005. Holocene processes. Earth Surface Processes and Landforms 21: solifluction, climate variation and fire in a subarctic 67–76. landscape at Pippokangas, Finnish Lapland, based on

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp Solifluction Processes, Dovrefjell, Norway 47

radiocarbon-dated buried charcoal. Journal of Qua- Washburn AL. 1967. Instrumental observations ternary Science 20: 533–548. of mass-wasting in the Mesters Vig district, Ripley AG. 2004. An investigation of soil structure Northeast Greenland. Meddelelser om Grønland interaction resulting from the freezing and thawing 166: 318 pp. of slopes. Unpublished PhD thesis, University of Washburn AL. 1979. Geocryology: a Survey of Periglacial Dundee. Processes and Environments. Edward Arnold: London. Smith DJ. 1988. Rates and controls of soil movement on a Washburn AL. 1999. A high Arctic frost-creep/gelifluc- solifluction slope in the Mt Rae area, Canadian Rocky tion slope, 1981–89: Resolute Bay, Cornwallis Island, Mountains. Zeitschrift fu¨r Geomorphologie, Supple- Northwest Territories, Canada. Permafrost and Peri- mentband 71: 25–44. glacial Processes 10: 163–186.

Copyright # 2008 John Wiley & Sons, Ltd. Permafrost and Periglac. Process., 19: 31–47 (2008) DOI: 10.1002/ppp